Fetch the Correct Chain Head During AttestHead

[rauljordan]

This is part of #1565

---

Description

Write why you are making the changes in this pull request

When fetching the information necessary to attest to a beacon block, we need to fetch the chain head from the beacon node. Instead, we were fetching a block at the assigned slot for the attestation, which could be nil and is different than what the spec mentions.

Another problem is that sometimes attesters are assigned to the same slot as proposers in rare instances, and in those instances we should be prioritizing proposer assignments.

Additionally, there was a discrepancy in fetching the epoch and justified epoch boundary block roots. The reference issue in the spec is here

Write a summary of the changes you are making

This PR uses the ChainHead() function from the db to fetch the canonical head when it is time to attest. It also changes the logic of the if condition in assignments to prioritize proposers over attesters.

if v.assignment.ProposerSlot == slot {
	return pb.ValidatorRole_PROPOSER
} else if v.assignment.AttesterSlot == slot {
	return pb.ValidatorRole_ATTESTER
}

[codecov]

Codecov Report

Merging #1596 into master will not change coverage.
The diff coverage is n/a.

@@           Coverage Diff           @@
##           master    #1596   +/-   ##
=======================================
  Coverage   70.96%   70.96%           
=======================================
  Files          94       94           
  Lines        6695     6695           
=======================================
  Hits         4751     4751           
  Misses       1527     1527           
  Partials      417      417

[prestonvanloon]

Please add a comment here since this is now an important distinction. Something like: A validator could be assigned to propose and attest in the same slot. Proposing takes priority.

Also, if a validator is selected to do both, should they do both?

[rauljordan]

#1613

[terencechain]

LGTM